Switch

Multi-Agent Debate (MAD) is an emerging paradigm that leverages the reasoning abilities of Large Language Models (LLMs) by encouraging them to collaboratively solve problems through human-like discussions. However, current MAD methods typically constrain agents to follow fixed discussion pipelines, repeatedly applying the same discussion act for a predetermined number of rounds, which limits their effectiveness and adaptability in complex and diverse tasks. To address this limitation, we propose Analyze–Compose–Execute (ACE), a novel debate framework in which agents dynamically execute the discussion actions according to the dialogue context. By analyzing the current responses of agents, ACE selects appropriate acts from a predefined Atomic Discussion Acts Library (ADAL), which are composed into a discussion action to be executed in the next round, to enable truly dynamic debate. We conduct extensive experiments on the challenging benchmark Big-Bench Hard (BBH) benchmark. ACE achieves state-of-the- art results on 17 out of 23 tasks, with an average performance gain of 8.5% across all tasks, demonstrating the effectiveness and robustness of our approach.

<div class="section abstract"><div class="htmlview paragraph">The global shift to electric vehicles (EVs) is vital for reducing greenhouse gas emissions, but their sustainability hinges on effective battery lifecycle management. This review examines the interplay between Life Cycle Assessment (LCA) and circular economy (CE) principles in EVs, with a focus on both international trends and India-specific challenges. We analyze CE strategies such as extending battery lifespan, second-life applications, and recycling integrated with LCA to evaluate environmental impacts from raw material extraction to disposal. Key areas include battery chemistry, LCA methodologies, policy frameworks, and industrial practices, informed by a synthesis of over 50 peer-reviewed articles, technical papers, and sustainability reports. Challenges include inconsistent LCA baselines, low material recovery in informal recycling, and regulatory gaps, particularly in India. Despite these, innovations like solid-state batteries and advanced recycling techniques offer promise, potentially reducing emissions by 30–40 percent through closed-loop systems.</div><div class="htmlview paragraph">Research gaps remain in areas like the durability of recycled materials, economic viability of CE strategies, and socio-ethical considerations. This review provides a holistic overview, actionable insights, and a roadmap for integrating CE into EV design and policy, especially tailored to India’s evolving automotive ecosystem. By addressing these issues, it aims to guide policymakers, industry stakeholders, and researchers toward a more sustainable, circular future for transportation.</div></div>

Vulnerability detection across source and binary code is an important yet challenging task in software security research. Existing deep learning methods are mostly limited to a single modality and fail to effectively leverage complementary feature information between source and binary codes, thereby limiting detection capabilities and application scenarios. In this paper, we propose a cross-modal vulnerability detection framework based on modal adapters and memory alignment, <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{MA}^{2}$</tex>-VD. Specifically, we first address heterogeneity differences across modalities using modality-specific adapters. Then, we construct a cross-modal feature memory bank based on FAISS, which implicitly aligns features by dynamically retrieving matching cross-modal features during inference to achieve memory alignment. We evaluated <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{MA}^{2}$</tex>-VD on the dual-modal NVD dataset that contains both source and binary code samples. The results show that <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{MA}^{2}$</tex>-VD outperforms the existing baseline methods by <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$5.21 \%, 1.23 \%, 9.86 \%$</tex>, and 5.71 % in terms of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$F 1$</tex>-score, precision, recall, and accuracy, respectively. Ablation studies and incremental experiments further demonstrate the effectiveness of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{MA}^{2}$</tex>-VD in cross-modal vulnerability detection.

in de palliatieve eindelevens-en rouwzorg.Ze werkt hiervoor samen met sociale en gezondheidswetenschappers en organisaties die cultuur, literatuur, zorg en welzijn samenbrengen.

Reducing our society's energy demand is critical to address the sustainability challenge. While the Internet currently accounts for 1–-1.5% of global electricity consumption and continues to grow, the energy demands of one of its core components---routers---remain poorly understood. The available power data is limited and not fine-grained enough, offering little actionable insight into strategies for effectively reducing the Internet's energy consumption.

<title>Abstract</title> 26 September, 2025. Research Square has withdrawn this preprint due to significant concerns regarding the integrity of the work. Maintaining high standards of ethical conduct is essential to our platform.

White Rabbit has emerged as a promising technology for time-scale dissemination, providing resilience to satellite techniques. To achieve state-of-the-art synchronicity at the sub-nanosecond level in a White Rabbit network, a complete characterization of physical effects, components, and the reference time scale is required in order to remove systematic offsets not compensated by the White Rabbit protocol. Due to this complexity, only a few long-range White Rabbit networks can claim synchronicity to a national time scale on a sub-nanosecond level. We present here such a network with the vocation to disseminate the Swiss official time-scale UTC(CH). To that end, the White Rabbit network layer is multiplexed into two dark channels in the L-band of a deployed telecommunication fiber network covering nearly 500 km of fiber. We characterize its performance in terms of stability and accuracy and find an uncertainty of 289–415 ps. This validates its potential to disseminate UTC(CH) at the sub-nanosecond accuracy and precision level. The results confirm that White Rabbit provides the necessary performance to distribute and compare the national time-scale UTC(CH) over optical fiber networks at the national and eventually at international levels.

Over the past two decades, much progress has been made in the construction of nonlocal sets without entanglement. Recently, Li and Wang [Quantum 7, 1101 (2023)] proposed an open question that an orthogonal product set (OPS) $\mathcal{S}$ is locally stable with minimum cardinality in ${⨂}_{i=1}^{n}{\mathbb{C}}^{{d}_{i}}$, and the numerical lower bound for such a set was given by Wang et al. [Quantum Inf. Process. 22, 15 (2023)], $(|\mathcal{S}|\ensuremath{-}1)|\mathcal{S}|\ensuremath{\ge}{\ensuremath{\sum}}_{i=1}^{n}({d}_{i}^{2}\ensuremath{-}1)$, but the existence of the construction remains unresolved. In this work we construct locally stable sets with minimum cardinalities ${d}^{2}$ and ${d}^{2}+1$ via 1-orthogonal cyclic product sets in ${({\mathbb{C}}^{d})}^{\ensuremath{\bigotimes}{d}^{2}}$ for odd $d$ and ${({\mathbb{C}}^{d})}^{\ensuremath{\bigotimes}{d}^{2}+1}$ for even $d$, respectively, thereby partly addressing Li and Wang's open question and confirming the existence of the locally stable OPS with the lower bound. Furthermore, we present a locally stable orthogonal cyclic product set of size $n$ in ${({\mathbb{C}}^{d})}^{\ensuremath{\bigotimes}n}$ and a locally stable OPS with smaller size in general multipartite systems. Our results show a perfectly cyclic structure of locally stable OPSs and provide new insight into the role of local stability without entanglement in quantum information theory.

Intrusion detection systems (IDSs) are essential in Intelligent transportation system (ITS) for detecting and identifying malicious activities. Although deep learning is commonly used in IDS, its limited interpretability hinders large-scale deployment. Additionally, the computational and storage constraints of ITS devices pose challenges to traditional IDS solutions. To address these issues, we propose RK-IDS, an interpretable IDS model combining rough set (RS) theory and the Kolmogorov-Arnold Network (KAN). RS selects critical features without sacrificing detection accuracy, while KAN identifies patterns for precise classification. Performance metrics, including accuracy, precision, recall, false alarm, F1-score, and runtime, are derived from classification results. Ablation experiments using the CSE-CIC-IDS2018 and CIC-DDoS2019 datasets show that RK-IDS achieves accuracy of 95.20% and 96.88%, respectively. These results suggest RK-IDS offers better detection capabilities than traditional neural networks, making it highly applicable to modern ITS networks.